This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Symmetry equivalent low resolution reflections in a good quality macromolecular crystallography data set typically scale to an R-merge value of 2-4 % depending on data redundancy. However, simulations taking into account random and systematic errors arising from typical experimental set-ups predict a lower R-merge. A possible source of error not accounted for in simulations is pixel-to-pixel variation in the response of the detector. In the case of the CCD detectors in use at SSRL beam lines, such variation may arise because each optical fiber has a different efficiency but the point spread function blurs them all together in the flood field. However, sharp diffraction spots only average over a given set of fibers. To test this hypothesis, a very redundant partial data set was collected from the same crystal at two different detector positions. When merging the images from two different part of the detector the R-merge falls as the square root of the redundancy, as expected for random noise. However, when processing the data sets from different detector position separately, a systematic difference of 2.5 ? 5 % was found. The size of the error correlates with the spot size, which further suggests a pixel to pixel variation for the detector (i.e., when the error introduced by this effect decreases when averaged over a larger number of pixels). More experiments to determine the influence of this effect on the measurement of the anomalous signal in SAD and MAD experiments are planned to form the basis for software developments to mitigate any effects.